I just took a file to the outside and inside bearing races on my bearings. I could easily file both races and even the balls themselves. These are not hardened. The listing on Amazon said 52100 hardened steel… B.S.
I will be taking a file to the new set of bearings I get and checking hardness before installing. They don’t arrive for a few days though…
What did arrive today was two new steppers just to try… at this point I want to eliminate all possibilities… I am tired of this roller coaster ride “its fixed” “its not fixed”…
Installed them and the problem still exists… so the smoking gun is really pointing at bearings right now but I have been burned so many times there is no way I am going to say I am sure that’s it until this thing does 100 laps without a stumble!!!
And as always, while replacing the motors I learned more… I was looking at what was going to be involved in replacing the bearings and I realized that I had installed the upper bearing bolt from the outside towards the build area which meant that when the new bearings arrived I would have to take the stepper off again to get that bolt out to replace the bearing. So again, do it the way the instructions suggest… someone before you may have learned a thing or two that was important that you may one day figure out on your own… doh…
So bolts installed correctly in anticipation of replacement bearings… Hopefully they are better quality…
Well Amazon’s speed of delivery is beyond amazing… combine that with lack of impulse control and… nevermind…
New bearings arrived. I could not look “through” the seals from one side to the other. I tried a quick file test to see if I could file a corner of the race with a regular file… I will buy a new file now… these are hard…
I was pretty excited… The X and Y axis move more freely than it EVER did even when the bad bearings were new!!! Huge improvement.
Then I got X and Y axis back together and tested it and still had issues. Yup… Thunk thunk as it moved.
Here is a video I posted of my X axis just moving it with the LCD knob so we aren’t talking anything but controller… no GCODE involved… just moving it along the axis using the Move Axis LCD and the scroll wheel.
Core is not installed… so there is very little load.
Here is what I observed.
Initial test, X and Y using V510S firmware (RAMPS), X motors in series, Y motors in series. Y axis seemed fine but X axis was very infrequently missing a step. One point is that Y axis had the original stepper motors driving it, X axis had the two new steppers I ordered last weekend and tested in a previous post. Y are 84oxin and X are 63ozin. Frankly under ZERO load I would expect ANY motor to drive it without losing steps.
Flashed with V510D firmware and put motors back to each their own driver just to see if it made a difference. Worse with 510D and seperate drivers. Bot hX and Y missing steps.
Flashed with V509D firmware and same result as V510 firmware.
Desperation move… flashed V505D using the .HEX file thinking may my Visual Studio Code/Platformio/Marlin builder are messing up. I heard thunks when homing the thing… both axis thunking… And when using LCD to Move Axis in 10mm steps lots of thunks…
Back to pondering… I’ve got nothing at this point. Tapped for ideas.
Well. I have been building a lot of ZenXY’s lately and you know it sounds like to me a non-parallel build.
Measure your diagonals. That sounds like the tube slipping through the clamp because it is getting narrower and wider (like it does on a sloppy built zen).
Diagonals differ by less than 2mm and Y differs by 1mm end to end… which is disappointing because I was sure I had that better when I started (mainly because I was still waiting for parts and had WAY too much time to kill so I spent it making sure I had it square…
But…
To confirm if this was the issue I loosened one clamp on the Y axis. Tube was completely free to move in or out. That would completely eliminate binding due to out of square. Ran it up and down after zeroing and still thumped… (dimensions are the MPCNC defaults from the calculator BTW… I wanted it to be able to mill aluminum so I figured small and stiff was the best approach)
Then I thought it may still be skewed if my dual limit switches are skewed so I homed Y axis. Powered off the motors and let it “relax” and made sure there was no skew whatsoever between left and right side then powered on the motors and ran it back and forth some more, still thumping.
Oh and just to show the level of desperation, I went for even older firmware… 404… and just for yuks I tried 404 compiled using Visual Studio Code/Platformio and Arduino IDE…
No change… even right after power up when you use Auto Home on the LCD you hear thumping as it heads for 0,0 and this is till with one clamp loose on the axis and nothing skewed.
If you over tensioned the belt you could have broken the fibers inside, but it would thunk at the same place or be very noticeable in a pen drawing of something like a circle. Plotting a circle nearly as large as your build can handle might actually reveal some things. Do several to see if the issue repeats or randomly or not.
Loosen the belts see if it still thunks?
Rotate the tubes, maybe they have a rough spot?
Seriously I feel like you have changed everything and still have the issue.
Just hearing you say that at least makesw me feel less tupid that I haven’t got it working. Thx.
Apparently I tried a few things I didn’t mention in this blog, I mean thread I already tried loosening the belts to see if over tension was causing some sort of binding.
I have watched it go back and forth and there is no pattern to the missed steps that I can figure. At times I was holding a felt marker to the carriage and putting a mark on the pipes when I heard something to see if the length between marks was the circumference of something… for a while it was… the circumference of a bearing… but with all new bearings and things moving smooth as sild along the rails… no new patterns in the marks.
This morning I thought what about power… I have my original 12V 15A supply. I tried an ATX supply instead. Same result… This morning I thought maybe power… so I stuck 4000uf electrolytic where the 12V connects to the RAMPS 1.4 board… same result…
But one thing I haven’t tried since replacing the bearings is power the mega2560 from a seperate 12V supply from the steppers… Will try that tonight…
Oh and one other thought was to try my RAMPS board on my Acro Laser Engraver and see if that works fine there…
I’ve been following this thread with interest and I was wondering…
Just a thought, have you checked the surface of the tubing? (I’m assuming you have galvanized EMT, if not, disregard my comments). If it is galvanized tubing, maybe there are some high and low blobs of the galvanized finish causing bumps in the road, so to speak?
That is an excellent thought and a legitimate concern (next time I go stainless for this and other reasons).
As it turns out the machine had enough testing and operating that the galvanized EMT tubing has flaked off all the galvanizing where the bearings contact and the tubes now have excellent flat spots where the bearings contact and when I saw this I marked the tubes in case I had to move or remove any of them so I could reinstall them and take advantage of the nice smooth flat surfaces that have formed. Fortunately the crap bearings took any initial abuse and the new bearings are running on a nice smooth surface.
That reminds me about one thought I had about my old bearings. The picture I posted of the new vs old inner race showed big spalling in one are of the inner race and I said that looked like the bearing had done a lot of small motions in one area. I was totally out to lunch on that. In MPCNC all the bearing inner races are stationary. This means that all pressure in a bearing is applied to only one side of the inner race, the side that faced the surface that the outer race contacts. Explains the pattern… (but a good bearing would NOT do that).
So, learned something. That’s what keeps me moving forward.
It seemed to me that the motor furthest from the controller was missing more than the near one so I thought just maybe the 1m stepper motor wire plus the .75m extension to get to the controller were too long for such thin wire (I have no idea what gauge they use) so I made up an 18 gauge wire set for the far motor and tried that. No change.
It did seem like the looser the belts the less it missed so I thought maybe the idlers were suspect. Replaced them. No change.
How warm are your drivers getting? If not very, I would bump the current up little by little. It seems to me that your motors aren’t getting enough power. The DRV8825s support up to 2.5A, so if you have those available, use them so you will have a little more headroom. My guess is that your 84oz/in motors have a 2A max per coil, so increasing your current shouldn’t hurt.
Another thing you might try, and it’s probably nothing, but lowering your microstepping and recalculate your steps per mm, to see if it’s a command buffering issue. I mean, 16 or 32 microstepping at the speeds your testing with shouldn’t be a problem, but it’s something else to try.
I am using DRV8825 drivers (but tried also with A4988) and I am running them at I believe 1.6 amps and while upping the current is possible, I have a feeling that I would later run into issues as I load up the machine when actually milling…
But I will still try it to see…
I have tried reducing microstepping to 16 and that had no effect…
I have my DRV8825 drivers set to .88v which if I read the specs right puts me just over 1.6amps. Currently on the Y axis I have some 63ozin steppers that are rated for 1.6A so they should be able at that current…
The other day I said I changed my idlers… well I lied… I changed one pair on the side that sounded like it was losing steps… but having learned the hard way (see previous discussion of loose grub screw on motor pulley) I decided that was dumb… change all four on Y axis then retest…
Man was I excited… for a while… it did better but still loses steps and gets worse the longer you run it. Belts almost loose enough to skip teeth and still loses steps.
The rails slide along so smooth on the new bearings there is now way its a bearing binding issue…
Drivers are fan cooled and cool to touch so I seriously doubt an overheat issue on the drivers.
Tonight in a real “OMG I better check that” moment I took all four motor pulleys off and counted the teeth just to make sure I didn’t have some strange 16T 15T combo somehow… but if I had that I would expect after homing things would start off good and get worse with distance as things skew. Its still totally random where it loses a step.
Oh and on the random chance that my stepper wires were somehow inducing noise in my limit switch wires that was somehow making it back to the arduino and causing some sort of hiccup I homed the MPCNC then unplugged the limit switches completely at the RAMPS board end and ran the test and same issue.
How do I change the subject from Not Lost Steps to not lost steps initially but now it is!!!
Spare CNC shield came tonight. Going to program it with GRBL and install a set of A4988 drivers I have spare and see how that works. I tried this before but that was with the bad bearings… I am realizing I had so many issue in the past I have to go retry some of the things I tried before now that some issues are resolved…
Not really. Anything above 0.7v on the drv is extremely hot and needs active cooling (which you have) and running steppers at the rated max is pretty nuts. They will get extremely hot fast. I would find the current that keeps them under 50C sustained.
At the same time unless you have a real ramps board, that could very well be a crappy clone. Trust me I have seen it all with the thousands that passed through my hands.
Always hard to know if you have a quality RAMPS board or a crappy one…
I swapped to my second RAMPS/mega/DRV8825 setup to see how that worked. (i.e. all new controller driver hardware, only the LCD reused) and nothing moved… then I put my pulleys back on the motors…
Still hearing thunk… but then I got to thinking, am I hearing something that isn’t really there? Am I hearing missed steps or is that just a bearing noise or a belt noise?
So I edited the GCODE I was sending (which for this testing was a simple square 320x330mm which is close to my build plate size. I added a move to 300,300 then a pause and I marked my EMT where the trucks were, then I had it run around my square 50 times then moved back to 300,300 and pause and I re-marked my EMT to see where it thought 300,300 was after 50 laps.
I am not hearing things… it is missing steps.
The X axis was the best. First image below is the “near” motor. i.e. the motor with the shortest wires to the controller/driver. The second is the far motor. The “A” in the pictures is the “After” position and the left side of the mark was where the truck was. On both pics home (0,0) is to the left. You can see the near motor was as far as I could tell spot on. The far motor was out by quite a bit. Interesting to note that there was a cumulative effect… the “after” marks all ended up further from “home” than the before marks…
Below pics are the Y axis. On the first pic home is on the right, second pic home is on the left. First picture is the far motor and its out by 50mm second is the near motor and its out by 5mm.
I am rerunning my test right now with microstepping set to 8 instead of 32 on the Y axis just to see if that affects things. It affects my ears… sounds horrible with 1/8 stepping vs 1/32…
I should add that these tests were all no load with core removed so it is just the X and Y trucks moving…
From looking at the pictures could it be your belts be slipping on the pulleys? Your belt teeth look rather pointy. Normal GT2 belt teeth have a like a half-circle profile. Maybe it’s just shadows?
In some earlier testing I loosened the belts to the point where they would jump a tooth (sloppy loose) and that was a different sound than what I am having so I don’t think that’s it. But thanks for reading and responding.
I already tried loosening the belts to see if over tension was causing some sort of binding.
